A systems approach to the cellular and molecular organization of neural circuits for representation of space

One of the most challenging problems in science is to understand how the molecules expressed by nerve cells in the brain enable thoughts and actions to take place. This is of fundamental importance for understanding how brains work. It will also underpin future industrial development of therapies for neurological and psychiatric disorders, and of biologically inspired computing technologies. Synthesis of molecules and assembly of cells is similar in the brain and other organs of the body, but the brain is distinguished by its ability to efficiently perform computations of considerable complexity. These computations rely upon communication of electrical signals between nerve cells. Some important computations are carried out by groups of nerve cells organized into modules, but how these modules relate to organization of electrical signaling between nerve cells is not known. This is important because molecules that control electrical signaling are a critical molecular link between gene expression and cognitive processes.

We will focus on a sub-region of the brain called the entorhinal cortex. During exploration, nerve cells at the upper end of this region form a module that encodes an animal's location at a relatively high resolution of approximately 30 cm. Lower down within this region, different modules of nerve cells encode location at lower resolutions. As existing approaches rely on recording electrical activity from neurons in live animals it is currently exceptionally challenging to examine their physical basis. We aim to solve this problem by instead using in vitro experiments in combination with quantitative and predictive computational models.

We will first establish if electrical properties of single nerve cells or their connections have a modular organization. We will use electrodes to record from many nerve cells in single slices of tissue. If electrical properties contribute to modular organization, then we expect cells from the same network to be more similar to one another than cells from different networks. We will next evoke coordinated network activity while making electrical recordings simultaneously from four cells at a time. We expect to identify cells that are part of the same module by specific correlations in their activity. To identify molecules that organize electrical properties and connectivity, we will identify candidate genes that mark modules. We will then determine if they label specific subgroups of neurons based on their electrical properties and connectivity.

Data obtained at each stage of experimentation will guide development of computer models. By comparison of the experimental results with the model predictions we will be able to refine and improve the predictive power of the models, while also identifying functions that the model may not yet explain and that will therefore require further investigation. In this way we aim to reveal new computational principles for brain operation and to ultimately enable direct links to be established between gene expression, electrical signaling and brain function.

The models and experimental results generated will be of benefit and application in several areas. 1) By establishing basic links between genes, electrical signaling and computation by nerve cells, the study will be important for understanding the healthy brain. They will form a key foundation for further investigations of how specific genes influences brain function. 2) The brain region that we will focus on is an important target for drug discovery. The computational models that we build will enable dry lab testing of potential therapeutic strategies in development by pharmaceutical or biotechnology companies. 3) The principles uncovered may stimulate future design of biologically based computational devices. For example, to improve navigation by robots, and to develop neurally inspired architectures to improve the energy efficiency of computational hardware.

We propose to investigate the cellular and molecular substrates for computations based on organization of neurons into discrete modules. We will focus on a sub-region of the brain called the medial entorhinal cortex (MEC). Neurons in the dorsal MEC form a module that represents an animal's location through grid like firing fields with relatively high resolution. At increasingly ventral locations neurons form modules with progressively lower spatial resolution. This modular organization is thought to be of exceptional computational significance, but we lack the cellular and molecular understanding of modules required to address this experimentally.

The main objectives of the proposed work are to establish whether intrinsic neuronal properties or local circuit connections have a modular organization, to identify molecules that distinguish populations of MEC neurons with distinct and possibly modular properties, and to develop and test biophysically constrained computational models of modular grid firing. These models will be used to evaluate and predict the roles of specific circuit properties in modular computations. We will investigate substrates for modularity by recording intrinsic membrane properties and gamma frequency oscillatory activity from multiple neurons in single brain slices. We will use these in vitro assays to establish whether identified candidate molecules label neuronal populations with distinct intrinsic properties or connectivity.

Data obtained at each stage of our analysis will be used to refine and improve the predictive power of computational models of grid firing, while also identifying functions that models may not yet explain and that will therefore require further investigation. The proposed work will direct new understanding of relationships between gene expression, electrical signaling and neural computation, and will enable future experiments to establish the computational and cognitive roles of neuronal modules.

The proposed work will reveal fundamental principles for cellular and molecular organization of computations in a brain area that is critical for navigation, learning and memory, and that is believed to be of great importance for healthy aging and as a target for drug discovery. Potential beneficiaries range from technology and pharmaceutical industries in the commercial private sector, through researchers in applied fields, in particular those oriented towards lifelong health solutions, to the general public as a whole. We describe below benefits in each area of impact. In some cases the critical path to impact may be direct, for example by immediate application of our research outputs to commercial development of new technologies. In other cases it will be through application of our fundamental findings to further applied research either in industry or academia. In our separate Pathways to Impact statement we describe diverse activities that we will employ to facilitate impact in each area. Academic impact is outlined elsewhere in the proposal.

1. Alterations in neural circuits are believed to be central to cognitive changes that accompany aging, but our lack of understanding of how circuit organization enables neural computations is a substantial obstacle to establishing which aspects of the aging process one should focus in order to promote healthy aging. Because the entorhinal cortex is one of the brain areas believed to be most important for cognitive changes that accompany aging, our results will therefore be of benefit to a wide range of specialists and organizations with an interest in developing strategies to promote healthy aging.

2. It is widely recognized that the drug discovery process is severely hindered by a lack of basic understanding of the cellular and molecular organization of brain areas important to healthy neurological and mental function. Biotechnology and pharmaceutical industries will therefore benefit from the new knowledge and research tools generated by the project. For example, a molecular understanding of modular organization of brain circuits may enable new diagnostic and treatment strategies, while our detailed computational models may help design and predict effects of pharmacological tools.

3. Technology industries are increasingly applying biologically inspired solutions to commercially important problems. Areas that may benefit from the proposed research include development of self-navigating systems, for which it may be beneficial to incorporate principles used for computation in animals, and development of computer hardware, for which computation in the brain may serve as a model for more energy efficient technologies.

4. The proposed project will also contribute to UK capacity building in systems biology. This has been identified by the BBSRC as a strategic priority of long-term benefit to the UK. The proposed work will provide training for the postdoctoral research associates employed to work on the project and for PhD, Masters and undergraduate students who will have the opportunity to work on computational models and experimental systems that develop from the project. The University of Edinburgh is particularly well placed for the project to contribute to postgraduate training, with several successful PhD and Masters programs, both within the host School (Biomedical Sciences) and within the School of Informatics.

5. Understanding the brain is widely recognized as one of the most important challenges in modern science and public demand for knowledge of how our brains work is reflected in the high media profile given discoveries in neuroscience research. Because of their relevance to human cognition and aging, the results of the proposed study therefore have the potential to contribute to public engagement with systems biology and research into the brain.